In this paper we explore why judging a manager on their underlying skill is a better measure than purely focusing on their past investment returns.

Look at any investment document and you will see the caveat “Past performance is not a good predictor of future performance”. Regardless of the caveat, past performance continues to provide a comfort blanket for investors and as a result plays a larger part than it should when it comes to appointing or maintaining an investment manager. In the current climate of historically low interest rates, future market returns are expected to be low and in some cases negative, so it’s understandable that investors are looking for repeatability when it comes to out-performance.

However, to truly understand whether a manager is capable of continuing to perform we need to look further than total returns and assess the manager’s actual skill level. In favorable market conditions even a manager with poor skill can seem to do well, but of course a manager with good skill would expect to do better. In turn, non-favorable markets matched with poor manager skill can have a compounding, negative impact on total returns.

In this paper we explore why judging a manager on their underlying skill is a better measure than purely focusing on their past investment returns.

Past performance is a thing of the past

Global fixed income markets have been both a beneficiary and a victim of the low-inflation, low-growth environment seen since the Global Financial Crisis (GFC). Falling yields do indeed have their benefits, particularly for investors who already own high quality fixed income securities. Many investors simply do not care that 10-year US government bonds trade at a meagre 1.8% yield when total returns in 2016 have been in excess of 9%. Looking in the rear view mirror, government bond returns in developed markets look seductively good (mid-high single digits), particularly when compared to the anaemic cash rates on offer.

The reality is that expected future returns are indicated by current bond yields; and with yields at all-time lows, returns are expected to be lower. So naturally some investors will be looking for additional return, above the market.

When it comes to achieving above benchmark returns, clients have two key areas to achieve this: asset allocation and manager skill. Historically, the asset allocation decision was the critical decision, as the market return (beta) overwhelmed any manager’s additional contribution through their ability to outperform (alpha). But in the current world of increased quantitative-easing (QE), where rates are low and equity market returns hover in the mid-single digits, the contribution to total return from a manager’s alpha grows relative to the return on the market (or index). Furthermore, in the period following QE, the ability of a manager to outperform may be all that stands in the way of negative total returns. So with the acknowledgement that manager skill is more important in the current market, how does a manager’s repeatability to deliver out-performance fit into the manager selection process?

Stability of skill and opportunity dictates repeatability

Portfolio performance is a fairly complex composite of influences, most of which are unstable over time. Skill, for example, for any specific investment sector, comes and goes, sometimes disappearing and often disappointing for lengthy periods. Similarly, the level of available excess return - a close cousin of volatility - ebbs and flows, changing the opportunity available for even the most skilled decision-maker. For many portfolios, the universe of potential investments varies, which further complicates assessing repeatability. And many firms change personnel, change process, change methods, or don’t operate robust investment processes, contributing to the variability of portfolio outcomes.

In our view, repeatability is best considered from a purely statistical perspective. If we assume both the investment process (methodology and team) and the universe of potential (and available) investments are stable, then the question of repeatability boils down to stability of both skill and investment opportunity. To demonstrate we take the following extreme example; assume that a portfolio produces active return by investing, long or short, in a single return source i.e. taking one active stock position and otherwise holds the benchmark. If this were the case, the repeatability of the portfolio’s active return would boil down to the behavior of this stock price and the ability of the underlying process to generate stable skill in predicting this one stock’s behavior.

In this very simple case, assume that the manager’s skill is constant, year in and year out, then the only factor contributing to instability in the manager’s active return would be the instability of the stock itself. While there are a variety of measures available to quantify the stability of any alpha source, we choose a measure we call “opportunity”. Opportunity represents the amount of return available for a perfectly skilled analyst – we call this the perfect capture. This measure simply assumes an analyst could forecast perfectly the daily change in the price of an alpha source.

The chart below shows the yearly aggregation of daily changes in the level of US interest rates representing the yearly tally of opportunity.


Aggregation of daily changes in the level of US interest rates

Source: Colonial First State Global Asset Management, Investment Opinion Network (ION).

Now assume that the manager has a constant skill level of 10% – meaning they can capture 10% of each year’s opportunity. Looking at 2006, when there was 176 basis points (bps) of opportunity available, the manager would have captured 18 bps of return. In 2008 they would have captured 36 bps (10% of 356). The portfolio’s total active return would be the total capture in the period, amplified by the active exposure (overweight/underweight positioning), less the transaction costs. In very rough terms and using the above example, over the span of two years the active return has doubled (from 18 bps to 36 bps), despite a stable investment process and a constant level of investment skill (10%). Hence, even with constant skill from the manager, the capture opportunity available in the marketplace can swing significantly and hence produce very different return outcomes.

Whilst in the above example we have assumed a constant skill, the reality is that instability in skill is typically a much larger contributor to return variability than market opportunity. Skill, especially over short periods, can swing widely and is commonly the dominant reason managers fail to repeat their out-performance. In different market environments (such as 2006 and 2008) this can result in a wide range of returns despite the manager skill levels staying the same. This is most noticeable when the low skill level coincides with the high opportunity environment or when the high skill coincides with the low opportunity environment.

In the table we can see that when low skill is captured in a high opportunity environment coupled with high skill in a low opportunity environment then the overall skill captured is significantly lower (9 bps per annum) than if the high skill occurred in the high market coupled with low skill in the low market (45 bps per annum). Hence, constant skill levels over time can result in notably different alpha being achieved when we consider the changing market environment.

In addition, it’s important to acknowledge that the above example is the capture for only one alpha source in two market scenarios. This is obviously not a reflection of the global market, hence imagine the complexity once you add a couple hundred alpha sources. Portfolio returns are indeed complex composites of influences.

So how can we assess a manager’s ability to repeat their previous performance?

Perhaps the best way to assess repeatability is to break this question into its three component parts:

1. How stable is the manager’s skill?

Managers should be able to describe and explain their skill and the variability of their skill. There are a number of ways skill might be measured, but its measurement should be regular and a standard part of any investment process.

2. How stable is the opportunity set for the product?

Secondly, managers should be able to describe the universe of potential portfolio risks, and how this universe has changed over time. Changes in the opportunity to produce return isn’t a controllable item, but can have material influences on product return (as we saw in our above example). Looking just at the active return doesn’t give a clear indication of how the manager has performed from a skill perspective. As we demonstrated above, a manger’s skill may have increased while the total opportunity set decreased, leaving the portfolio with a lower active return or vice versa.

3. How stable is the process, and the people providing investment judgments?

The final piece is a standard discussion topic, and certainly an important question within the broader topic of manager repeatability. But it is only one piece, and arguably the piece most likely to deceive. Unless the manager understands the skill contribution of all of their investment team, when people leave it’s not clear whether good skill or poor skill is walking out the door. If the organization attempts to measure their skill, they likely understand how staff changes relate to the overall skill. If nothing else, skill measurement might mean the manager understands and values repeatability, and are doing their part to sustain their ability to repeat.

If an investor can collect appropriate answers to these three questions, they should have a reasonably good handle on a manager’s skill and hence whether a manager is likely to repeat prior performance. While these are perhaps unusual discussion topics, their importance is likely to grow as we navigate a world of low returns and diminished market volatility. In this new world, the question as to whether a manager is likely to repeat performance is perhaps as important as how to appropriately allocate any portfolio’s assets.

Please click here to download as a PDF.